Open Access Open Access  Restricted Access Subscription Access

Towards Autonomous MAV Control in Atmospheric Turbulence

S. Watkins


We have created a repeatable, controllable turbulent environment for the study of manmade MAVs and perhaps birds and the larger flying insects. By reconfiguring a large industrial wind tunnel, and using a selection of
grids and screens, we produced a range of turbulence intensities and integral length scales and considered that these conditions cover the
majority of turbulent flow conditions relevant to MAVs flying in a typical range of wind speeds in complex terrain. Our studies are restricted to well mixed turbulent flow, rather than investigating discreet gusts and local
effects (such as might be found in extremely close proximity to buildings etc). The work described here forms part of a larger program which is focused on determining the influence of MAV type (fixed wing, rotary and flapping) on ability to fly in relatively turbulent flow conditions. Manned craft have received much attention with respect to the effects of atmospheric turbulence but generally this is clear-air turbulence at high altitude.

Here we provide an overview of the flight test facility and several series of flight tests with fixed and rotary wing craft. We also detail initial flight tests of a fixed-wing aircraft under a variety of constraints designed to limit the
degrees-of-freedom to the test craft. Two purpose-designed rigs are described that; 1) permit the craft to perform pure roll, at adjustable angles of attack, and; 2) permit pitch and heave with minimal roll, lateral and foreand-aft translations. These rigs were used to simplify the tuning of an autopilot system that could include six DOF inertial inputs, pressure sensing (via wing dynamic pressure taps) and strain measurements from spars.

The results demonstrated that with no autopilot enabled control was extremely challenging, resulting in some terminal departures. However, relatively steady flight was achieved with a degree of autopilot input via roll rate tracking and damping. A series of relatively small, rapid inputs to the ailerons (at higher frequencies than could be achieved by an experienced human pilot) greatly assisted overall controllability. Furthermore the system demonstrated an ability to hold reasonably straight and level flight in relatively homogenous turbulence (approximately 8% turbulence intensity with 1m length scale) and when more discreet gusts were generated via upstream bluff bodies. Subsidiary measurements are described that include pressure, force and moment measurements on wings held fixed in turbulent flows. These help explain why the pressure-based autopilot system did not perform as well as the inertial

Full Text:




  • There are currently no refbacks.